List your company
ResearchFormerly WizWand

Mathematical Reasoning on In-Distribution Avg

45.6Average ScoreTRAPO
13.46421.80730.1538.493Dec 15, 2025Jan 1, 2026Jan 19, 2026Feb 6, 2026Feb 24, 2026Mar 14, 2026Apr 1, 2026
Updated 7mo ago

Evaluation results

MethodLinks
TRAPOTraining Paradigm=Semi-supervised, Labeled Samples Count=4K, Unlabeled Samples Count=12K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1245.6
Fully SupervisedTraining Paradigm=Supervised, Labeled Samples Count=45K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1245.5
HiLLBackbone=Qwen2.5-7B-Instruct2026.0444.2
Fully SupervisedTraining Paradigm=Supervised, Labeled Samples Count=4K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1243.1
HiLLw/o TWBackbone=Qwen2.5-7B-Instruct2026.0442.7
TRAPOTraining Paradigm=Semi-supervised, Labeled Samples Count=1K, Unlabeled Samples Count=3K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1242.6
SAGEBackbone=Qwen2.5-7B-Instruct2026.0442.3
LUFFYBackbone=Qwen2.5-7B-Instruct2026.0441.7
GRPOBackbone=Qwen2.5-7B-Instruct2026.0441.1
Scaf-GRPOBackbone=Qwen2.5-7B-Instruct2026.0441
Self-certaintyTraining Paradigm=Semi-supervised, Labeled Samples Count=1K, Unlabeled Samples Count=3K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1240
Token-level EntropyTraining Paradigm=Semi-supervised, Labeled Samples Count=1K, Unlabeled Samples Count=3K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1240
Sentence-level EntropyTraining Paradigm=Semi-supervised, Labeled Samples Count=1K, Unlabeled Samples Count=3K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1239.7
Fully SupervisedTraining Paradigm=Supervised, Labeled Samples Count=1K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1239.4
TTRLTraining Paradigm=Semi-supervised, Labeled Samples Count=1K, Unlabeled Samples Count=3K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1239.2
Self-certaintyTraining Paradigm=Unsupervised, Unlabeled Samples Count=45K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1238.3
TTRLTraining Paradigm=Unsupervised, Unlabeled Samples Count=45K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1238.2
BaseBackbone=Qwen2.5-7B-Instruct2026.0437.8
Qwen-InstructTraining Paradigm=Original, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1237.6
Token-level EntropyTraining Paradigm=Unsupervised, Unlabeled Samples Count=45K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1237.6
Sentence-level EntropyTraining Paradigm=Unsupervised, Unlabeled Samples Count=45K, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1232.6
HiLLBackbone=Llama-3.2-3B-Instruct2026.0424.6
SAGEBackbone=Llama-3.2-3B-Instruct2026.0423.9
HiLLw/o TWBackbone=Llama-3.2-3B-Instruct2026.0423.7
GRPOBackbone=Llama-3.2-3B-Instruct2026.0421.9
Scaf-GRPOBackbone=Llama-3.2-3B-Instruct2026.0421.5
Qwen-BaseTraining Paradigm=Original, Backbone Model=Qwen2.5-Math-7B, Sampling Temperature (T=0.6)=0.62025.1219
BaseBackbone=Llama-3.2-3B-Instruct2026.0417.8
LUFFYBackbone=Llama-3.2-3B-Instruct2026.0414.7